Can hybrid search boost retrieval performance in a non-English, highly specialized domain like law? In my final PhD paper (COLING 2025), I investigate this question by combining various retrieval techniques across two realistic scenarios: 1️⃣ zero-shot, where we assume **no**

AccessPath using AI to analyze and make physical spaces more accessible built a custom deep research pipeline using cohere, Tavily, Jina AI

How jina do vibe-check embeddings? Here we go github.com/jina-ai/correl…

Is it another win of Bayes theorem? The EM looks like to add some conditional distribution shift to the prior knowledge (pretrained model).

We've got some Research Scientist Intern spots open in the Bay Area Jina AI . Hit me up if you're into Embeddings, Rerankers, or small language models for better search stuff. Would be awesome if you've worked with long context before, or have experience with

We released a new model: jina-embeddings-v4 - an universal embedding model for multimodal and multilingual retrieval. - trained on a wider scope than DSE, ColPali, etc. - support MRL, late interaction, etc. 🤗huggingface.co/jinaai/jina-em… jina.ai/news/jina-embe…

Submodular optimization for token/sentence selection from long contexts. Here's an interesting exp: first used jina-embeddings-v4's multi-vector feature to extract token-level embeddings from a passage, then applied submodular optimization to cherry-pick the tokens that provide

How Jina AI built its 100-billion-token web grounding system with Cloud Run GPUs Google Cloud cloud.google.com/blog/products/…

Meet my colleague Michael Günther at SIGIR 25 if you are interesting in embedding/reranking/small language model training

All you need is right context selection! Jina AI introduced a simple classic methodology for selection task.

jina-reranker-m0-GGUF is here huggingface.co/jinaai/jina-re…

vLLM is finally supporting our multi-modal reranker jina-reranker-m0 huggingface.co/jinaai/jina-re… This is neat!

Congrats! contextual chunk embedding is undervalued. It is good to see more and more players push it forward after jina-v3 later chunking.

wow it is super fast ⚡️

The hero behind Qwen3, GSPO go brrr

New benchmark drops: JinaVDR (Visual Document Retrieval) evals how good retrieval models handle real-world visual documents on 95 tasks in 20 langs—think layouts packed with graphs, charts, tables, text, images. We're talking scanned docs, screenshots, the works. JinaVDR pairs

The best paper I have read this year! Key takeaway: ... may reflect a fundamental shift in the era of LLMs, where underfitting becomes less problematic than overfitting.

We are at Qdrant 's Vector Space Day 🚀 in Berlin on Sep 26. We'll talk about "Vision-Language Models: A New Architecture for Multi-Modal Embedding Models" and also share some insights and learnings we gained while training jina-embeddings-v4. 🎫 lu.ma/p7w9uqtz

Just curious, could any guys tell me how Instruction following reranker help in your app?

Today we're releasing jina-code-embeddings, a new suite of code embedding models in two sizes—0.5B and 1.5B parameters—along with 1~4bit GGUF quantizations for both. Built on latest code generation LLMs, these models achieve SOTA retrieval performance despite their compact size.